The last great recession was due to precisely this sort of spending pattern plus a collapse in payment. Banks may be healthy, for now, but they can't keep lending forever with no recover. This is not a good sign.

I mean maybe, in theory, but how often do prices actually drop on hotels? Prices are usually based on occupancy, so unless somebody cancels a block of a hundred rooms or something, this seems unlikely to make a meaningful difference in hotel revenue.

Meanwhile, if this had happened five years ago, it would have meant not going on trips for me, because traveling with my dad in his last couple of years had a decent risk of having to cancel.

There are people for whom being able to cancel a trip without significant penalty is the only thing that makes travel possible. And if hotels are more concerned with maybe losing a few bucks in rare circumstances than they are with whether the elderly and disabled get to have a vacation at all, then F**K those hotels. They aren't worth doing business with.

His surname is one transposition away from "AI Mode".

Anthropic's entire pitch has always been safety. Innovation like this tends to favor a very few companies, and it leaves behind a whole pile of losers that also had to spend ridiculous amounts of capital in the hopes of catching the next wave. If you bet on the winning company you make a pile of money, if you pick one of the losers then the capital you invested evaporates. Anthropic has positioned itself as OpenAI, except with safeguards, and that could very well be the formula that wins the jackpot. Historically, litigation and government sponsorship have been instrumental in picking winners.

However, as things currently stand, Anthropic is unlikely to win on technical merits over its competition. So Dario's entire job as a CEO is basically to get the government involved. If he can create enough doubt about the people that are currently making decisions in AI circles that the government gets involved, either directly through government investment, or indirectly through legislation, then his firm has a chance at grabbing the brass ring. That's not to say that he is wrong, he might even be sincere. It is just that it isn't surprising that his pitch is that AI has the potential to be wildly dangerous and we need to think about safety. That's essentially the only path that makes his firm a viable long term player.

I'm banking on that

The talk about space being cold, having more light, etc. is pretty much nonsense. More like it has less regulations and bigger budgets. However, I could imagine some kind of low latency compute being needed for applications like:
- Luna: Automated robotic exploration and construction drones wanting 2 second lag, different countries are working on it
- Asteroid mining robots: Onboard compute, or lower latency needed on arrival / when opposite Sun
- Telescope on other side of the sun: Maybe not needed
- Solar system / Oort exploration robots: Maybe needed years from now
- Deep space exploration robots: Maybe needed years from now

Just an armchair guy but it seems like:
- Setting up in-system relays alone should be sufficient for delivering AI+Human based commands when low latency is sufficient, i.e. in transit. And will be required for operation in planetary shadow. We have something at Mars, and for the Far Side of the Moon we are apparently working on Lunar Communications Relay and Navigation Systems (LCRNS) for Artemis.

- Moving compute nearby for lower latency (1 second) will be needed when robots actually touch rock or enter planetary shadows, if we want them to move with any kind of speed. For the Moon, libration points would give 400ms round trip which is probably enough, or in-orbit is better.

- There will be a lot of advancement in processor capability/size and to a much lesser extent space propulsion over the years robots are traveling the deep dark, so it may be better to wait as long as possible so that a miniaturized, low power, high compute package can be delivered on a fast rocket to be there when they need it.

Yeah, because all even/all odd is (from basic statistics) rare, and happens to be rarer than the percentage of people who play all-even or all-odd, so you'd be more likely to split any winnings.

It's more like "spell check my email" I expect. Probably not a lot of people actually pasting spreadsheets into context. I am guessing 100% of non-native and overseas users are using it to help them write English, at least based on one company I know (I think if they have a contract they figure it is safe...).

And as for myself I write a lot of documents and email in a second language. I am very careful not to post anything sensitive but have found Claude to be amazingly good at checking emails or installation guides I write in a different language which I write well (I have been a pro translator in the other direction even), but as a non-native actually still learn better writing style from Claude, in addition to finding typos or better words. It is to the point that translating as a job must be in dire straits. For non-technical users I am guessing once you drink the kool-aid you are going to slide towards pasting anything.

Batteries are catching up faster than it will be cost-effective to build nuclear in the US. A month ago, Bremen Airport announced they had integrated a new sodium-ion battery with a 400 kW output and 1 MWh capacity into its infrastructure. The entire thing apparently fits in roughly one twenty-foot shipping container, and there is almost certainly room to expand that to additional batteries to provide power through the night and beyond.

Beyond that, Peak Energy just signed a deal to build up to 4.7 GWh of sodium-ion batteries by the end of the decade. This follows a successful 3.5 MWh demo project in Colorado. Time will tell if they can successfully scale up and avoid the fate of Natron energy, which just ceased operations.

But the market does appear to be moving rapidly in the direction of battery storage regardless of individual solutions, with BNEF forecasting another 92 GW of output and 247 GWh of capacity just for batteries in 2026, almost a quarter more than 2025. They expect growth of 2 TW/7.3 TWh by 2035. Some people think that's conservative, similar to how solar has blown past everyone's expectations from even 2015. I think if the iron- and vanadium-based flow battery demos work as hoped, that could let cheap grid-level battery installations soar beyond anyone's expectations. Whether lithium-ion, sodium-ion, or flow, they will land far sooner than we could build equivalent nuclear plants. It will be better to greatly expand solar, like over parking lots, irrigation canals, and other places where they can lower heat and supply energy to the batteries. It's politically easier and can provide more jobs in more areas that don't require college degrees. Many more winners than sticking with nuclear or fossil fuels.

Modern models will take into account statistical biases in lotteries. For example, here's ChatGPT:

Pick good lottery numbers for the Virginia powerball lottery

Good question. First, a reality check: Powerball is pure chance — there’s no “good” numbers that are more likely to win. Every set of 5 white balls (1–69) + 1 Powerball (1–26) has exactly the same probability.
wwwuat.valottery.com
+1

That said, I can give you a few tips + a sample “random but balanced” pick, if you’d like to pick your own numbers rather than use Quick Pick.

Some tips when choosing your Powerball numbers

Spread your numbers
Try to pick a mix across the range (low, mid, high). According to lottery-strategy advice, not clustering all your numbers in one group is better.
samlotto.com

Avoid patterns
Many people pick birthdays, anniversaries, or sequential numbers. That’s fine—but if you win, you might have to share with more people because these are popular picks.
lotteryvalley.com

Use “due” or “hot” numbers carefully
Some sites (like lottery-analysis tools) track which numbers are “hot” (drawn recently) or “overdue” (haven’t appeared in a while). But remember: past draws don’t change future odds.
m.lottostrategies.com

Mix even and odd
According to some statistical analyses, it’s less likely to draw all odd or all even numbers.
samlotto.com

Play responsibly
Odds of winning the Powerball jackpot are extremely long: 1 in ~292 million.
lotterycritic.com

Sample “balanced” Powerball pick

Here’s a random-but-balanced combination you could use just for fun (or as a basis to tweak):

White balls: 8, 23, 41, 57, 65

Powerball: 12

If you like, I can generate 5 different “smart random” picks for the next Virginia Powerball draw (balanced by probability and number distribution). Do you want me to do that?

It's a damned shame '99942 Apothis' will miss us in a few years

It's not too late to fix that.

The thing is, it very much *won't* be random numbers. It's about as far from a random number generator as you can get.

In a really bad model, one would expect it to likely be "numbers it's seen the most in the context of a lottery". But in a good model, esp. a thinking model, one would expect it to think over which sorts of numbers are statistically over-chosen (birthdates, etc) and avoid them in giving its answers.

The funny thing is, think of how this woman will interact with ChatGPT from now on. ChatGPT could say "Abraham Lincoln was married to John Wilkes Booth", and she'd be thinking, "Okay, that sounds really, really wrong, but on the other hand, it picked the right lottery numbers, so..."

Ideally, it should be a hair space (because em dashes in web fonts are borderline illegible without it), but Slashdot does not support Unicode, and   gets silently swallowed by Slashdot's HTML parser. Besides, we all know that AP style is the one true style, and it demands space.

No, an en dash is used for numerical ranges and certain compound hyphentions, not for parentheticals.

I'll go with NHTSA and NASA over the "Barr Group" ambulance chasers, thank you. Barr found that it's possible if you get like a cosmic ray to flip just the right bit you could stick the throttle on (but still not make it overpower the brakes). NHTSA and NASA investigated not just the software but the actual cases. In not a single actual case that they investigated did they find that it wasn't well explained by either stuck pedals or pedal misapplication (mainly the latter).